#Bom Currículo - ATS Resume Builder

#UNDER CONSTRUCTION

Build ATS-friendly resumes using AI.

#About

Bom Currículo is an open-source project that generates professional resumes optimized for Applicant Tracking Systems (ATS).

The system can use multiple optional sources of information, including:

Current resume
LinkedIn Profile (PDF export)
GitHub profile
Portfolio
Personal information
Target job description

The user does not need to provide every source. The only requirement is enough information to generate a resume.

#Features

LinkedIn profile analysis
GitHub profile analysis
Portfolio analysis
Job description matching
ATS-friendly resume generation
Resume optimization

#Screen prototype

An initial interface prototype was also developed in Figma to validate the user experience, feature organization, and the platform's main flow before final implementation.

The screen is still evolving, but it already represents the first visual proposal of the user dashboard, including:

ATS score overview;
General resume performance indicator;
AI-powered optimization tips;
List of uploaded resumes;
Application progress tracking;
Sidebar menu with access to the platform’s main areas.

Note: the design is still under development and may be adjusted as the product evolves.

Project screens preview

#Roadmap

Resume generation
LinkedIn PDF parser
GitHub integration
Portfolio analysis
Job matching
Multiple templates
Export to PDF
Export to DOCX

#Contributing

Contributions are welcome!

Please read CONTRIBUTING.md before submitting a Pull Request.

#Running the project

Create the environment files:

cp .env.example .env
cp backend/.env.example backend/.env
cp frontend/.env.example frontend/.env
cp bot/.env.example bot/.env

On Windows PowerShell:

Copy-Item .env.example .env
Copy-Item backend/.env.example backend/.env
Copy-Item frontend/.env.example frontend/.env
Copy-Item bot/.env.example bot/.env

docker compose up --build

The Docker stack includes Ollama and automatically downloads the model defined by OLLAMA_MODEL (default: qwen3:4b) into the persistent ollama_data volume. The first startup can take several minutes because that model is about 2.5 GB. The bot reaches Ollama through the Docker network at http://ollama:11434; it does not use the bot container's localhost.

The local default favors CPU-only development. Resume requests disable Qwen's thinking mode, cap generated output at 2048 tokens, and stop the complete AI provider chain after 240 seconds. Laravel waits up to 260 seconds for the bot, while Nginx waits up to 280 seconds, allowing a bot timeout to reach the client as a structured response. On a Ryzen 7 5700G, a representative dense resume completed with qwen3:4b in about 209 seconds; slower CPU-only machines may still need a faster external provider or an asynchronous workflow.

To confirm that the model is available or download it again:

docker compose exec ollama ollama list
docker compose exec ollama ollama pull qwen3:4b

Set OLLAMA_MODEL in the root .env before starting the stack to use another model. OLLAMA_BASE_URL may point the bot at an external Ollama deployment instead, but its address must be reachable from inside the bot container. For better local quality on sufficiently fast hardware, set OLLAMA_MODEL=qwen3:8b; that model is about 5.2 GB and remains fully supported.

Services:

Health endpoints:

Backend: http://localhost:9000/up
Bot: http://localhost:8000/health

#License

This project is licensed under the MIT License.

Instalar en iPhone/iPad

Bom Curriculo